NPS Surveys: Design, Sampling, and Acting on the Score

What NPS Actually Measures (And Doesn't)

Net Promoter Score is the most widely tracked customer metric in B2B and consumer products. It's also one of the most misunderstood. Companies report NPS like it's a moral verdict, follow up on it inconsistently, and use the free-text reasons to make product decisions they shouldn't.

The honest reality: NPS is a useful tracking metric. The score itself — on the standard "How likely are you to recommend [product] to a colleague?" 0–10 question — correlates well with retention and revenue growth across industries. The free-text reasons are weaker signal than the score, because customers articulate what they consciously think they want, not what they actually do.

This guide covers how to design, sample, score, and act on NPS correctly. It pairs with the broader customer feedback loop discipline.

How NPS Is Calculated

The formula:

1. Ask "How likely are you to recommend [product] to a colleague?" on a 0–10 scale
2. Categorize responses:
   • Promoters: 9–10
   • Passives: 7–8
   • Detractors: 0–6
3. NPS = % Promoters − % Detractors

NPS ranges from −100 (everyone is a detractor) to +100 (everyone is a promoter). The score is reported as a number, not a percentage — "an NPS of 42," not "42% NPS."

What's a Good NPS Score?

Industry / Segment	Good	Great	Best-in-class
B2B SaaS (mid-market)	30+	50+	70+
B2B SaaS (SMB)	20+	40+	60+
B2B SaaS (enterprise)	35+	55+	70+
Consumer SaaS	30+	50+	70+
E-commerce	30+	50+	70+
Banking / insurance	10+	30+	50+
Healthcare	10+	30+	60+
Telecom / cable	-10+	10+	30+

Industry benchmarks matter. An NPS of 45 is excellent for telecom, mediocre for B2B SaaS. Don't chase a universal target; chase movement against your own baseline within your industry.

How to Design the NPS Survey

The standard survey has two questions:

1. "On a scale of 0–10, how likely are you to recommend [product] to a colleague?"
2. "What's the primary reason for your score?" (optional free text)

A few important design choices:

Use the Exact Wording

"Recommend to a colleague/friend" is the validated phrasing. Don't rewrite it to "rate your experience" or "satisfaction" — those measure different things and break comparability to industry benchmarks.

Keep It Two Questions

The temptation to bolt on more questions ("What features do you want?", "How is our support?") destroys response rates. Keep the survey to two questions. Run additional research separately.

One Reminder, No More

Send the survey once, then send one reminder 3–4 days later to non-respondents. Beyond that, you accumulate annoyance without meaningful response lift.

Embed In Product or Email — Don't Use a Pop-Up

In-app embed (e.g., a small bar at the top of the dashboard) or transactional email both work well. Modal pop-ups are increasingly blocked or dismissed reflexively.

When to Send the NPS Survey

Timing is the most-skipped variable. NPS sent at the wrong moment produces misleading data.

Moment	Should you send?
Immediately after signup	No — too early
30 days after onboarding completion	Yes — primary baseline
Every 90 days after that	Yes — quarterly cadence
Immediately after support resolution	Use CSAT, not NPS
Immediately after billing	No — confounds the data
During an outage or known incident	No — pause survey delivery
After a major feature launch	Yes, but tag the cohort separately
To users who haven't logged in for 30+ days	Yes — captures disengagement signal
To customers giving notice of cancellation	Skip — use a churn interview instead

The pattern: send during steady-state usage, not during exceptional moments (positive or negative). Send to a representative sample, not just engaged users.

How to Sample for NPS

Most companies survey all active customers. For small companies (under 200 customers) this is fine. Above that, sample strategically.

Customer Segment	Sampling Approach
All paying customers	Survey all if <500 customers
Paying customers, segmented by tier	Stratified sample: 100% of enterprise, 25% of SMB
Free / trial users	Sample 10–20% to avoid noise
Recently churned (within 60 days)	Skip — use churn interviews instead
Long-tenured customers (>2 years)	Always survey — they're your expansion base
New customers (<30 days)	Skip — too early for stable signal

A response rate of 25–40% is typical for in-product surveys; 10–20% for email surveys. Below 15%, you have non-response bias problems (only engaged customers respond, inflating your score). Above 50%, you might be over-surveying and damaging engagement.

How to Read NPS Results

Look at Trends, Not Snapshots

A single quarter's NPS is noise. Look at trends over 3+ quarters. A score moving from 38 → 42 → 47 over a year is meaningful; a one-time score of 47 is barely actionable.

Segment by Tier

Aggregated NPS hides important segment dynamics. Enterprise customers might be at +60 while SMB customers are at +25 — those are different businesses requiring different responses. Always segment.

Watch the Detractor %, Not Just the Average

Two companies with the same NPS of 30 can have very different distributions. Company A: 50% promoters, 30% passives, 20% detractors. Company B: 40% promoters, 50% passives, 10% detractors. Same NPS, very different health. Company B is more resilient; Company A is more bipolar.

The detractor percentage is the leading indicator for churn. Above 20%, focus on detractor remediation.

Free-Text Reasons Are Lower Signal Than You Think

The free-text "why" responses are useful for tagging themes but unreliable as roadmap input. Customers articulate consciously held opinions, which often don't match their actual behavior. Use the reasons to inform interview questions, not to drive feature decisions directly.

How to Act on NPS Results

The Detractor Follow-Up

Detractors (0–6 scores) are the single highest-leverage segment. A personal outreach within 48 hours of their score can convert 20–35% of detractors back to passives or promoters.

Pattern:

1. Personal email from the founder, CSM, or account manager
2. Reference the score and free-text reason
3. Ask one specific question: "What would have to change to make this a 9 or 10?"
4. Schedule a 15-minute call if they want to talk

This is not a sales conversation; it's a service-recovery conversation. Roughly 50% of detractors who get a personal outreach respond. Roughly 35–50% of those who respond improve their experience materially.

The Passive Outreach

Passives (7–8) are the largest segment for most products and the hardest to act on. They're not unhappy but not enthusiastic. The lever: convert their conditional satisfaction into specific advocacy.

Pattern: ask passives what would make their score a 9. The answers reveal product gaps and feature opportunities more reliably than detractor reasons (because passives are engaged enough to articulate aspirations).

The Promoter Activation

Promoters (9–10) are your latent advocacy base. Most companies waste them by treating "high NPS" as the end of the conversation. Three high-ROI promoter activations:

• Reference program: ask promoters to be sales references for prospects
• Case study program: ask promoters to participate in a written or video case study
• Referral program: invite promoters into your referral program explicitly; promoters convert at 3–5x the rate of average customers

Failing to act on promoters is the single most-common NPS mistake. The score tells you they're warm; you have to make the ask to convert that warmth.

NPS Cadence and Reporting

Cadence	Activity
Daily	Track new responses; flag detractor scores for immediate follow-up
Weekly	Detractor outreach completed; passive follow-up triaged
Monthly	Score reported to product/CS team; theme tagging on responses
Quarterly	Full NPS report including segmented scores, trends, comparisons; published to investors and team
Annually	Benchmark against industry; review survey design and timing

Common NPS Mistakes

Treating NPS as a Vanity Metric

Reporting NPS in the board deck without doing detractor outreach is theatre. The score is only useful if it drives behavior — detractor recovery, promoter activation, segment-level action.

Adding Questions to the Survey

Two questions only. Every additional question reduces response rate by 10–20%. Run other research separately.

Surveying Too Often

Quarterly is the right cadence for steady-state. Monthly produces survey fatigue and inflates response variance.

Surveying During Confounding Events

Sending NPS during an outage produces a meaningless score. Sending immediately after a price increase produces an artificially low score. Pause surveys during major events and resume 30+ days after.

Comparing NPS Across Industries

A B2B SaaS NPS of 45 is good; a consumer banking NPS of 45 is excellent. Benchmark within your industry, not against the universal "Apple has an NPS of 72" headline.

Skipping the Personal Detractor Outreach

The highest-ROI activity in the entire NPS program. Skipping it because "we can't follow up at scale" leaves the most valuable signal on the table.

When NPS Isn't the Right Metric (Not For You)

Skip NPS in favor of other metrics if:

• You have under 50 customers. Statistical variance overwhelms any score signal. Talk to every customer individually instead.
• Your product is transactional, not relational. For one-off transactions (an event, a one-time purchase), CSAT (customer satisfaction) at the transaction level is more useful than NPS at the customer level.
• You serve regulated industries with mandated satisfaction frameworks. Healthcare and government often have their own measurement requirements that supersede NPS.
• You're in pre-product-market fit. NPS is noise pre-PMF. Use customer interviews and product analytics instead.

Conclusion

NPS works when treated as a tracking metric, paired with disciplined detractor recovery and promoter activation. It fails when reported in a dashboard without action.

Design the survey to two questions. Send 30 days post-onboarding and quarterly thereafter. Sample strategically. Segment results by customer tier. Follow up with detractors within 48 hours. Activate promoters into references, case studies, and referrals. Pair NPS with the broader customer feedback loop and retention strategies — and the score becomes a steering signal, not a decoration.

Frequently Asked Questions

What is a good NPS score for B2B SaaS?+

30+ is good, 50+ is great, 70+ is best-in-class for mid-market B2B SaaS. SMB SaaS benchmarks are 10 points lower (20+ good); enterprise benchmarks are 5 points higher (35+ good). Industry context matters — a B2B SaaS score of 30 is healthy; a consumer SaaS score of 30 is mediocre.

How often should I send NPS surveys?+

Quarterly is the right cadence for steady-state customers. Send the first survey 30 days after onboarding completion, then every 90 days. Don't send monthly — it produces survey fatigue and inflates response variance. Don't send less than quarterly — trends become impossible to detect.

What's the difference between NPS and CSAT?+

NPS measures overall relationship quality with a long-term frame ('Would you recommend?'). CSAT measures satisfaction with a specific transaction or interaction ('How satisfied were you with this support ticket?'). Use NPS quarterly for relationship health; use CSAT immediately after transactions for tactical feedback.

Should I act on the free-text reasons in NPS responses?+

Tag them for themes but don't drive product roadmap directly from them. Customers articulate conscious opinions that often don't match actual behavior. Use NPS reasons as inputs to deeper research (customer interviews, behavioral analytics), not as the primary signal for what to build.

How should I follow up with NPS detractors?+

Within 48 hours of their score, send a personal email from the founder or account manager. Reference their free-text reason. Ask 'What would have to change to make this a 9 or 10?' and offer a 15-minute call. Personal outreach converts 20–35% of detractors back to passives or promoters — the single highest-ROI activity in the NPS program.

Is NPS still relevant in 2026?+

Yes, with caveats. The score still correlates well with retention and revenue growth across most industries. The free-text reasons are weaker signal than they used to be perceived as. NPS is most useful as a quarterly trend metric paired with detractor recovery and promoter activation, not as a one-off vanity report.

Should I survey all customers or just a sample?+

Survey all paying customers if you have under 500 of them. Above 500, sample strategically: 100% of enterprise tier, smaller stratified samples of SMB. Skip recently churned customers (use churn interviews instead) and very new customers (under 30 days). Sampling reduces survey fatigue while preserving statistical validity.